box service keeps restarting, SyntaxError: Unexpected token '.'
-
I've been having odd issues with my Cloudron for the past week or so. Apps were running fine, but backup tasks were failing. It'd refer me to the logs, which when checked, stated the file had been moved or did not exist.
I'm set up to backup to S3 via rsync. It was working fine initially, so I'm not sure what happened. In any case, I created a snapshot (which supposedly saves state, and yet...) and made the boneheaded decision to reboot my VPS. It came back up, but I got the dreaded red "Cloudron is offline. Reconnecting..." banner. Something I've seen way too many times.
I attempted to restore via snapshot, still getting the banner. I'm honestly wondering if the Cloudron distribution that Hostinger uses is fully stable. This has to be my 3rd or 4th attempt. It'll work fine at first, but inevitably the ticking time bomb goes off, and all I get is a red banner for no good reason.
Cloudron-troubleshoot instructed me to run cloudron --fix-docker, as somehow Docker was a newer version. I've learned my lesson about running apt upgrade long ago, so I have no idea how that's possible. Would simply adding the PPA upgrade it as well?
Regardless, it appeared it was going to succeed, but when it tries to start box, it just keeps restarting. This leads to an output of
[FAIL] box service keeps restarting, please investigate the error by inspecting /home/yellowtent/platformdata/logs/box.log. The box.log (and 1, 2, and 3) are massive, and I have no idea what I'm looking for.Running
journalctl -xeu boxgave me the following:░░ Subject: Unit process exited ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ An ExecStart= process belonging to unit box.service has exited. ░░ ░░ The process' exit code is 'exited' and its exit status is 1. Jan 10 06:37:24 c.example.com systemd[1]: box.service: Failed with result 'exit-code'. ░░ Subject: Unit failed ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ The unit box.service has entered the 'failed' state with result 'exit-code'. Jan 10 06:37:24 c.example.com systemd[1]: box.service: Scheduled restart job, restart counter is at 2815. ░░ Subject: Automatic restarting of a unit has been scheduled ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ Automatic restarting of the unit box.service has been scheduled, as the result for ░░ the configured Restart= setting for the unit. Jan 10 06:37:24 c.example.com systemd[1]: Stopped Cloudron Admin. ░░ Subject: A stop job for unit box.service has finished ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ A stop job for unit box.service has finished. ░░ ░░ The job identifier is 868014 and the job result is done. Jan 10 06:37:24 c.example.com systemd[1]: Started Cloudron Admin. ░░ Subject: A start job for unit box.service has finished successfully ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ A start job for unit box.service has finished successfully. ░░ ░░ The job identifier is 868014. Jan 10 06:37:24 c.example.com box.js[514142]: /home/yellowtent/box/src/ldapserver.js:206 Jan 10 06:37:24 c.example.com box.js[514142]: if (req.app.accessRestriction?.groups?.indexOf(group.id) === -1) continue; Jan 10 06:37:24 c.example.com box.js[514142]: ^ Jan 10 06:37:24 c.example.com box.js[514142]: SyntaxError: Unexpected token '.' Jan 10 06:37:24 c.example.com box.js[514142]: at wrapSafe (internal/modules/cjs/loader.js:915:16) Jan 10 06:37:24 c.example.com box.js[514142]: at Module._compile (internal/modules/cjs/loader.js:963:27) Jan 10 06:37:24 c.example.com box.js[514142]: at Object.Module._extensions..js (internal/modules/cjs/loader.js:1027:10) Jan 10 06:37:24 c.example.com box.js[514142]: at Module.load (internal/modules/cjs/loader.js:863:32) Jan 10 06:37:24 c.example.com box.js[514142]: at Function.Module._load (internal/modules/cjs/loader.js:708:14) Jan 10 06:37:24 c.example.com box.js[514142]: at Module.require (internal/modules/cjs/loader.js:887:19) Jan 10 06:37:24 c.example.com box.js[514142]: at require (internal/modules/cjs/helpers.js:85:18) Jan 10 06:37:24 c.example.com box.js[514142]: at Object.<anonymous> (/home/yellowtent/box/box.js:7:18) Jan 10 06:37:24 c.example.com box.js[514142]: at Module._compile (internal/modules/cjs/loader.js:999:30) Jan 10 06:37:24 c.example.com box.js[514142]: at Object.Module._extensions..js (internal/modules/cjs/loader.js:1027:10) Jan 10 06:37:24 c.example.com systemd[1]: box.service: Main process exited, code=exited, status=1/FAILURE ░░ Subject: Unit process exited ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ An ExecStart= process belonging to unit box.service has exited. ░░ ░░ The process' exit code is 'exited' and its exit status is 1. Jan 10 06:37:24 c.example.com systemd[1]: box.service: Failed with result 'exit-code'. ░░ Subject: Unit failed ░░ Defined-By: systemd ░░ Support: http://www.ubuntu.com/support ░░ ░░ The unit box.service has entered the 'failed' state with result 'exit-code'. Jan 10 06:37:24 c.example.com systemd[1]: box.service: Scheduled restart job, restart counter is at 2816.SyntaxError: Unexpected token '.'appears to be the issue, but from what I've read the error only occurs in very old (<14) Node versions. Would really appreciate some assistance getting this up and running.If I have to nuke this box again, I think I'll forego Cloudron and just use the dedicated Nextcloud image. Far more limiting than I'd like, but better than endless failures.
~ SunDevil ~
-
J joseph marked this topic as a question
-
@SunDevil that is interesting indeed. I wonder how that is possible given that yellowtent is actually a disabled yser (you cannot do
su - yellowtent). Cloudron has also never used that specific version of node, so I am not sure where it's coming from.Can you debug further with
sudo -u yellowtent which nodeetc and remove that node? Maybeapt remove nodejs? The system node /usr/bin/node should be a symlink to /usr/local/node-20.18.0/bin/node .sudo -u yellowtent bashwill give you a bash with yellowtent user and then you can check where that node is coming from as well. -
@SunDevil said in box service keeps restarting, SyntaxError: Unexpected token '.':
learned my lesson about running apt upgrade long ago, so I have no idea how that's possible.
Sounds like your snapshots are all using whatever borked-up state the system was at when it was made. So, of course any restores to these snapshots are going to continue to not work. If you can indeed reinstall fresh, do that. I'm not familiar with how Hostinger offers its VPSes, but those extra node versions, Docker, etc., all got installed "somehow". If it isn't Hostinger...
-
@scooke Of course you're right. The only conclusion I can come up with is that simply adding the Docker PPA results in an
apt updateANDapt upgrade. The update is visible, but there's no indication it's runningapt upgrade.Node is more complicated. I'll double check, but I don't recall installing fnm or anything. I may have to reach out to Hostinger about these issues.
-
@girish I'm even more confused. To the best of my knowledge, I haven't messed with the Node version.
It's using
/usr/bin/node, which oddly enough seems to be the only package not symlinked. The first command you gave me shows Node v20.18.0. The second command shows Node v12.22.9. The latter appears to be a default package included with Ubuntu 24.04.1 LTS.Full Data:
Listing of /usr/bin, scoped to
grep nodeListing of installed packages, scoped to
grep node -
@SunDevil Perhaps Hostinger provides "non-fresh" servers to its customers with a variety of programs already installed to save you time. Normally a fresh server would not come with Docker, node, etc., at all. Cloudron needs the fresh server because it needs to follow it's own install script. Sounds like Hostinger has maybe gotten the two "fresh" settings overlapped.
-
@SunDevil it seems there is a whole bunch of node modules installed via apt on the system. Cloudron doesn't install these. Cloudron installs the node version it wants under /usr/local/node-20.18.0/ and then creates a symlink. Something like this:
apt remove nodejs ln -sf /usr/local/node-20.18.0/bin/node /usr/bin/nodeAfter this, node should be 20.18.0 . Then, you can
systemctl restart boxand it will start with the correct node. -
I checked out Hostinger... its Cloudron image can be installed AFTER the initial server is provisioned. One reason for the errors @SunDevil is having is if they installed the Cloudron image after provisioning the server, and the server had been "managed" prior to the Cloudron installation. I don't think asking them to run normal troubleshooting is going to work here with their experience level. I recommend @SunDevil get more direct professional help to try to undo whatever was previously installed, even if that is paid-help from the Cloudron team. I don't think they'll be better off just going the Nextcloud route as, if the same conditions persist/repeat, Nextcloud is also going to start acting wonky and good luck getting THAT solved! Paying for ssh-access-level-help is your best way forward and let a techy person clear out all that extraneous stuff. Good luck.
-
I'm starting to think giving Dokploy a spin may be worth it. Sounds more complicated than Cloudron, but then Cloudron is unstable.
This is either the third or fourth time I've lost my entire Nextcloud. It's just not worth the hassle with Cloudron.
That said, if you have any insights given my reply, please let me know. It seems to me this is unlikely to be rectified. If that's the case I'd prefer to just take the hit now rather than spend days troubleshooting to no avail.
-
@scooke I appreciate the honesty. I probably should reach out to Hostinger. Unfortunately, I think paid help is outside the question.
My problem. Well, I suppose if Dokploy or the regular Nextcloud image fails as well, then I'll just let my sub lapse.
It's either something I'm doing or it's Hostinger. Maybe I just don't belong anywhere near a VPS...
-
Not sure how this is related to Nextcloud, as all apps run within their own sandbox and are not related do nodejs versions on the host. Maybe to come back to the nodejs problem, did you run those node commands within an app container or the host system?
-
@SunDevil I'll keep sharing... I've tried Nextcloud on several different platforms, including JUST Nextcloud. It ALWAYS conks out on me. Always. So, my 2 cents, you'll have the same issue on Dokploy.
EDIT: Have you even checked out Dokploy??? I just watched their intro video. They install their software on your VPS from their servers! What?? Sorry, I'm not letting some third party have that access to my data. You create your ssh key on their system!!! You have to create a db separate from your app!!! Good luck keeping track of all those deets. Holy smokes, the number of steps needed to get this thing running is exactly why most of us have come to Cloudron. This Dokploy is so at the opposite spectrum of Cloudron. I can't imagine anyone thinking it will be "better".
